Students can receive adaptive, low-bandwidth math practice without a scarce one-to-one tutor.
Collections: Human still decides
Executive brief
A narrow, scheduled supplement can expand individualized practice without pretending to replace teachers.
AI value · Mathematics growth score
VerifiedTreatment growth was significantly higher with effect size 0.36 (p<0.001), described as roughly an extra year of learning.
Preliminary year-one evaluation; authors advise judicious interpretation.
Before
Attends normal math classes and studies with available static materials or limited human support. → Provides class-level feedback within available time.
After
Uses a school-provided phone for two weekly WhatsApp tutoring sessions during study hall. → Rori provides individualized practice and feedback while normal instruction continues unchanged.
Human boundary
Teachers and schools control curriculum, access, and assessment; students interact within scheduled sessions.
Why it matters
Students can receive adaptive, low-bandwidth math practice without a scarce one-to-one tutor.
Before
How the work ran before the change.
Student
Attends normal math classes and studies with available static materials or limited human support.
ControlSchool curriculum.
Teacher
Provides class-level feedback within available time.
ControlInstructional judgment.
What changed
Students can receive adaptive, low-bandwidth math practice without a scarce one-to-one tutor.
Decision rightHuman moves from creator to judge
After
How the same work runs now.
Student
Uses a school-provided phone for two weekly WhatsApp tutoring sessions during study hall.
ControlScheduled school setting and common curriculum.
Rori and teacher
Rori provides individualized practice and feedback while normal instruction continues unchanged.
ControlTeachers and schools retain curriculum and safeguarding authority.
Exception path
Connectivity, comprehension, or safeguarding concerns return to teachers and normal school support.
Decision authority
Teachers and schools control curriculum, access, and assessment; students interact within scheduled sessions.
| # | Actor | Action | Control |
|---|---|---|---|
| 01 | Student | Attends normal math classes and studies with available static materials or limited human support. | School curriculum. |
| 02 | Teacher | Provides class-level feedback within available time. | Instructional judgment. |
| # | Actor | Action | Control |
|---|---|---|---|
| 01 | Student | Uses a school-provided phone for two weekly WhatsApp tutoring sessions during study hall. | Scheduled school setting and common curriculum. |
| 02 | Rori and teacher | Rori provides individualized practice and feedback while normal instruction continues unchanged. | Teachers and schools retain curriculum and safeguarding authority. |
Work that left the path
Human role before
Teachers supplied instruction and limited individualized practice.
Human role after
Teachers retain instruction and oversight while AI adds scalable supplemental practice.
AI role
Decision mode: instructional tutoring. Selects and responds to math practice in chat; it does not grade the course or replace the teacher.
Control students continued identical classes and classroom hours without Rori access.Treatment growth was significantly higher with effect size 0.36 (p<0.001), described as roughly an extra year of learning.
Year-one intervention. · Approximately 500 students in grades 3-9 in Ghana.
Preliminary year-one evaluation; authors advise judicious interpretation.
One-to-one human tutoring is scarce and costly in the setting.Approximately US$5 per student.
Study estimate. · Basic phones on low-bandwidth networks.
Marginal cost excludes broader school implementation and device-management costs.
Anti-pattern
Generalizing one-year results to unsupervised home use or all subjects.
Questions
Portability conditions
Reputation risk
medium